From: VAFadeyev@lbl.gov
Date: Mon Oct 20 2003 - 01:12:23 PDT
Hi Ariel,
it is good that the results are consistent.
Your observation of the biases is very interesting. Was it ever checked before that the LC parameter
extraction is self-consistent and bias-free, say
for lower redshift cases?
The only reason why I switched off the randomization
was to save time. Otherwise I'd have to rerun
the code multiple times for every observation
sequence, derive the corresponding distributions etc.
Initially I did briefly checked that with randomization turned on there were no big biases
for z=1.3. This is close to your case of z=1.4,
where you don't have substantial biases either.
So perhaps this was not a very strict check.
Am rerunning that example for z=1.2 now. Will
report the results shortly.
Cheers,
vitaliy
----- Original Message -----
From: Ariel Goobar <ariel@physto.se>
Date: Monday, October 20, 2003 0:40 am
Subject: Re: Optimization of observations for the HST search
> Hi Vitaliy,
>
> your results seem consistant with what Jakob, Vallery and I have
> been looking at in Stockholm. However, we have also randomized
> the points according to measurement error and noticed that, not only
> is the width of the fitted stretch about 0.1 but the distribution
> of fitted lightcurves is also biased by a comparable amount. The
> stretch
> and date of maximum are tightly correlated, see attached figure.
> The one thing that is not 100% "right" in our simulations is
> that we did not use *exactly* the same template in the simulation
> and SNminuit.
> Ariel
>
> --
> ___________________________________________________________________
> Ariel Goobar (www.physto.se/~ariel)
> Department of Physics, Stockholm University
> AlbaNova University Center, SE-106 91 Stockholm, SWEDEN
> tel: +46 8 55378659 fax: +46 8 55378601
>
This archive was generated by hypermail 2.1.4 : Mon Oct 20 2003 - 01:14:41 PDT